Background: Critical care is one of the most resource-intensive and costly components of hospital care. Substantial variability exists across VA hospitals in the use of critical care for patients with acute respiratory failure (ARF), suggesting a knowledge gap about when and how to best use the intensive care unit (ICU). Furthermore, overuse of the ICU for patients who do not need critical care contributes to rising healthcare costs and may cause harm. Significance: First, this proposal will help clarify whether ICU bed supply influences ICU bed utilization. This has important implications for healthcare delivery across the VHA, as it would indicate that decreasing ICU bed supply may help prevent unnecessary ICU use without sacrificing quality of care, thereby improving the overall value of critical care. Second, it will identify modifiable organizational factors and care processes that enable the provision of high-quality and efficient care outside the ICU. Third, it will lead to the development of a hospital- level intervention that will tailor care and guide triage strategies for Veterans with ARF. Innovation: For this study, Dr. Vranas has created a novel dataset that combines highly-granular data from over 100 geographically diverse hospitals within the VHA. This proposal is also innovative in its application of Human- Centered Design Thinking to develop scalable interventions to improve the value of critical care at VA. Specific Aims: 1) Evaluate the association of ICU bed supply with ICU admission, and the association of ICU admission with outcomes; 2) Use a positive-deviance approach to identify organizational factors and care processes associated with high-value care of patients with ARF; 3) Use Human-Centered Design Thinking to develop and pilot test an intervention that will guide ICU utilization and improve the value of ICU care. Methods: Aims 1 and 2 will use an explanatory sequential mixed methods approach to evaluate organizational factors and care processes associated with high value ICU care. In Aim 1a, Dr. Vranas will perform patient-level multivariable regression analyses to test the hypothesis that increased ICU bed supply is associated with increased odds of ICU admission. To test the hypothesis that ICU admission is not independently associated with improved outcomes (Aim 1b), Dr. Vranas will use propensity score techniques to evaluate the association of ICU utilization with outcomes including mortality and discharge to home. In Aim 2, Dr. Vranas will use a positive deviance approach to identify hospitals with high/low ICU utilization, and above/below average risk-adjusted 30- day mortality for patients with ARF. Using purposive sampling, she will conduct medical ethnography and semi- structured interviews of key stakeholders at approximately 12 select hospitals. She will use Rapid Qualitative Inquiry methods to characterize processes associated with outcomes observed in Aim 1. She will also utilize the Consolidated Framework for Implementation Research to identify common domains that are modifiable and generalizable across healthcare settings. In Aim 3, Dr. Vranas will assemble a stakeholder taskforce and use Human-Centered Design Thinking to inform the development of an intervention designed to guide triage decisions for patients with ARF. She will pilot this intervention at the Portland VA hospital to evaluate its acceptability/feasibility, and to generate preliminary data for a future multicenter trial. Next Steps: This work will lead to the submission of multiple IIR proposals, including a hybrid effectiveness- implementation clustered randomized controlled trial of the refined intervention across multiple VA sites.